An operating cadence is not a meeting schedule. It is the rhythm through which evidence becomes a decision, a decision becomes owned work and the company checks whether that work changed the outcome.
When teams of 5 to 30 people feel chaotic, the reflex is often to add meetings. That creates more opportunities to describe work without improving how work moves.
A better cadence gives each forum one job. It also gives every recurring event a cancellation rule.
Download the one-page startup operating cadence and adapt the owners, inputs and timeboxes to your team.
The Four Speeds
Daily: protect flow
The daily mechanism is an exception check, not a round-robin status meeting.
Input only work that is blocked, at risk or affecting a customer now. For each item, answer:
- What outcome is at risk?
- What evidence shows the block?
- Who owns the next move?
- When will the next signal be visible?
If the exception queue is empty, cancel. Routine progress belongs in the work system.
For a remote team, this can be asynchronous. Publish the exception record by an agreed hour, let owners respond in the record and call only when a cross-functional decision is required. The broader remote startup operations playbook covers the communication architecture around this.
GitLab’s public handbook-usage model shows the operational alternative to keeping context inside meetings: documentation is treated as the shared source that people update through a visible contribution process. A small startup needs far less documentation, but the same separation helps. Stable context belongs in the record; live time belongs to unresolved judgment.
Weekly: make operating decisions
The weekly operating review should connect outcomes, queues and choices. It is not a deck recital.
Use a short pre-read containing:
- three to seven outcome measures with definition and owner;
- material changes since the previous review;
- the oldest or highest-impact waiting work;
- open risks and exceptions;
- prior decisions due for verification;
- decisions required this week.
During the meeting, spend little time reporting green status. Use the time for differences in interpretation and trade-offs. Record every decision with owner, due date, evidence and revisit condition.
If the team cannot name the decision that came from a discussion, it was a conversation, not an operating review.
Monthly: change the system
Weekly reviews help the team operate the current system. Monthly reviews decide whether the system itself should change.
Bring:
- repeated exceptions;
- handoff wait trends;
- incidents and overdue preventive actions;
- capacity compared with demand;
- procedures whose actual use differs from the document;
- customer evidence that challenges the current promise or workflow.
Choose at most one or two mechanism changes. Assign a verification date. A monthly review that creates ten process initiatives usually creates no process change.
Quarterly: reconnect operations to strategy
Strategy changes the work the system must support. The quarterly review asks:
- Which customer promise matters now?
- Which workflows became strategically critical?
- Which decisions should remain concentrated or move closer to the work?
- Which metrics no longer represent the outcome?
- Which ritual, report or tool can be retired?
The official Scrum Guide frames its recurring events as formal opportunities to inspect and adapt, and warns that skipping them loses those opportunities. You do not need Scrum events across the company. You do need a known moment when evidence can change the plan and a known record of what changed.
The Cadence Contract
Give each forum seven fields.
1. Purpose
One sentence describing the decision class. “Review operations” is too vague. “Resolve cross-functional risks to this month’s customer outcomes” is useful.
2. Inputs
Name the records required before the event. A dashboard link is not enough if nobody can explain the metric definition, date range or owner.
3. Output
State what the event must produce: decisions, changed priorities, assigned actions, accepted risks or no change with reasoning.
4. Owner
The owner protects the purpose, quality of input and closure of outputs. The most senior participant need not own every forum.
5. Participants
Invite people who supply evidence, make the decision or own the consequence. Everybody else can read the record.
6. Timebox
The timebox forces preparation and prioritisation. It is not a command to stop a consequential decision halfway through. If sessions routinely exceed it, narrow the purpose or improve the pre-read.
7. Cancellation rule
Cancel when no decision is required, the required evidence is absent or the queue is empty. Cancellation is evidence that the mechanism is protecting time, not evidence that the cadence has failed.
A Practical Week for a 12-Person Startup
This is a hypothetical example.
Every weekday, 09:45: owners update the exception queue asynchronously. A 10-minute call occurs only if two functions must resolve the same block.
Monday, 16:00: product, commercial and delivery owners publish their operating pre-read.
Tuesday, 11:00: 45-minute weekly operating review. The group decides on resource or customer trade-offs and checks last week’s decisions.
Friday, 15:00: teams close local work and update the evidence attached to decisions. There is no company-wide status meeting.
First Tuesday of the month: 60-minute system review focused on repeat exceptions, incident actions and flow data.
This is one configuration, not a benchmark. Customer support, regulated operations or multi-time-zone teams may need different daily mechanisms. Preserve the separation of purposes even when the calendar changes.
Measure the Cadence Itself
Every month, inspect:
- decisions produced per forum;
- percentage with named owner and verification date;
- actions closed without outcome evidence;
- repeated topics that indicate an unresolved rule;
- attendee hours;
- forums cancelled under their stated rule;
- important surprises that the cadence failed to surface.
Do not optimise for decisions per hour alone. Some weeks should produce a deliberate decision to preserve the current course. The quality test is whether the record shows the evidence and reasoning.
Common Failure Modes
Status theatre
People read information that could have been written. Move reporting to the pre-read and reserve live time for judgment.
Founder routing
Every agenda item becomes a question for the founder. Pair the cadence with a decision-rights matrix.
Ritual accumulation
New meetings are added but old ones never retire. Make retirement a quarterly agenda item.
Action without verification
Tasks are marked done, but nobody checks whether the operating signal improved. Give each material decision an expected evidence date.
Emergency contamination
Weekly reviews are repeatedly consumed by today’s incidents. Strengthen the daily exception path and write down how incidents become verified system changes.
Start with the downloadable cadence, run it for four weeks and change it from observed evidence. If your current calendar is full but work still waits between owners, bring the cadence and one stubborn queue to a working session.
